Rankings

Real-world LLM rankings, built from production traffic

Most leaderboards are benchmark runs. Nemo Router's rankings are aggregated from real customer traffic — what actually gets shipped, what costs what, what fails when. Live data, hourly refresh, anonymized at the source.

Notify me at launch See the model catalog

rankings · preview

Sample preview

#1 throughputgemini-2.5-flash

#1 latency p50142 ms

#1 cost / 1M tok$0.075

#1 reliability (24h)99.99%

Models tracked20+

AnonymizedReal trafficHourly

Live models: 97+
Providers: 77
Ranking categories: 6
Update cadence: Hourly

Sample data only. Live rankings ship once the network has aggregated enough traffic to be statistically meaningful — currently 97+ models on Anthropic, Google & OpenAI, with Anthropic (Claude), Google (Gemini), and OpenAI (GPT, o-series) live today — AWS Bedrock shipping next.

What ships at launch

Six rankings, all driven by real traffic

Rankings only matter when they reflect production reality. Each ranking below ships with the same methodology page — open dataset definitions, refresh cadence, and statistical caveats.

Token volume leaderboard

Top 20 models by rolling 30-day token throughput across the network, anonymized per-org — what teams actually ship with, not what they pinned for a benchmark run. Filter by capability tag; sparklines show 7-day momentum.

Performance matrix

Sortable matrix combining speed (TTFT, tokens/sec), quality (independent eval scores + community votes), cost efficiency ($/1M tokens), and reliability (success rate, error class). Pin a model to compare against the rest.

Provider market share

Live treemap of provider token share with weekly deltas and a 6-month historical mode — watch where customer traffic shifts after a price drop or a new model release.

Cost efficiency

Dollar-per-million-tokens across the catalog, normalized for input/output mix and tool-use overhead — provider list price plus observed effective price. The ranking that should drive routing-strategy decisions.

Trending models

7-day sparklines showing which models are gaining or losing share in token volume and active orgs — the early-signal page for which provider to evaluate next.

Latency benchmarks

P50/P95/P99 per model and provider, refreshed hourly from production traffic — streaming TTFT and non-streaming end-to-end, filterable by region, with outlier flagging when P99 spikes.

Methodology

Where the numbers come from

The integrity of the ranking is the integrity of the data source. Here is exactly what we use.

Anonymized network traffic

All rankings are aggregated from real production calls flowing through Nemo Router. We never log prompt content for ranking purposes — only aggregate counts, latency, cost, and provider/model identity.

Live cost headers

Cost data comes from the x-nemo-request-cost header on every call — provider-authoritative, never estimated. The same number that hits your credit ledger drives the ranking.

Public eval suites

Quality scores supplement throughput rankings using public eval results (HumanEval, MMLU, GSM8K, MT-Bench). We disclose every source on the methodology page.

Providers tracked

The shortlist that drives the rankings

OpenAI

Anthropic

Google

Gemini

Mistral

Get the launch email — first to see live rankings

One email when the data crosses the statistical-significance threshold. Then nothing until something interesting happens.

Notify me See the catalog